Thermal Management with Conductive Adhesives

Specifically, the hybrid packages fit the TO format (transistor outline packages). TO packages TO-3, TO-5, TO-18 and TO-220 have been analyzed. The TO-220 case study focused on a transfer molded plastic IC package (instead of metal cans) which used non-electrically conductive epoxy; while the other case studies focused on silver filled epoxy used inside the hybrid TO cans.[1,2,3 ]

Experimental Discussion

Case study #1 demonstrated the thermal resistance of silver epoxy versus solder for 20 mil x 20 mil chips fixed to the TO-18. The thermal resistance of junction-to-case (θJc) was determined. By knowing the bond-line thickness (BLT) and the chip dimensional area, the thermal resistance results have been back calculated into bulk thermal conductivity values. Specifically, the eutectic die-attach demonstrated about 2 °C cooler improvement over the epoxy die-attached chip. The soldered chip had a measured thermal resistance between 4.8 - 5.3 °C/W, while the epoxied chip was 6.7 - 7.0 °C/W. Using back-calculations, the BULK thermal conductivity of the solder was discovered to be about 41 W/m °K, while the epoxied chip demonstrated about 29 W/m °K.

The effect of BLT and chip area on thermal resistance measurements has also been demonstrated. The chip size varied from (20 mil x 20 mil) up to (120 mil x 140 mil). The results demonstrate that the larger the chip, the smaller the thermal resistance, which was expected. The effect on BLT was demonstrated from 0.5 mils to 5 mils, and the results indicated that the lowest thermal resistance has been attained on the smallest BLT, which was also expected.

Case Study #2 is very much like Case Study #1, however it used the TO-5 package. In this case, the size of the chip is 75 mil x 80 mil with a 0.5 mil BLT. The thermal resistance junction-to-case (θJc) has been measured similar to Case Study #1. Again, the thermal resistance of the soldered chip was about 2 °C cooler improvement over the epoxied chip. θJc for the soldered chip was 9.0 - 10.3 °C/W, while the epoxied chip was 11.3 - 12.6 °C/W. Having back-calculated the BULK thermal conductivity, this study indicated that solder had about 0.36 W/m °K while the epoxy demonstrated 0.27 W/m °K.

Case Study #3, is similar to Case Study #2, but it used a TO-3 package. The θJc was established for 7 silver filled commercially available epoxies and compared to soft solder controls. The die were bipolar transistors from Motorola; 2N-3055 chips which were 120 mil x 140 mil. A minimum of five samples were examined for each epoxy. The average BLT was calculated for each epoxy. The θJc for solder controls was 0.90 °C/W. The best epoxy tested showed a θJc of 0.99 °C/W, while the worst epoxy showed 4.9 °C/W.[1]

After back-calculating the BULK thermal conductivity, the solder control was detected to be 2.56 W/m °K, and the best epoxy value was 1.14 W/m °K. The worst epoxy tested was observed to be 0.35 W/m °K. The results of this study are somewhat contradictory to Case Study #1. It can be concluded that the package design of Case Study #1 was far better constructed than this example. The former resulted with solder having a BULK thermal conductivity of 41 W/m °K, while the latter had 2.56 W/m °K. Comparing the epoxy attached chip in Case Study #1 and #3, the former had 29 W/m °K, while the latter had 1.14 W/m °K.

Case study #4 is similar to #3 above in that it used the same TO-3 package and the same chips, however, this study observed the thermal resistance as a function of 150 °C/1000 hr. burn-in, as well as varied the effect of BLT. Six commercially available silver-filled epoxies were examined against soft-solder controls. θJc observations were made with the help of IR Thermograph; the model was Agema Thermovision 782. A color coded temperature gradient, accurate to 0.1 °C, was developed and recorded on the thermogram.[3]

θJc measurements were observed to be the lowest for the solder chip, which was expected, and a resultant value of 0.3 °C/W. The highest θJc obtained was 1.45 °C /W for an epoxied chip. Back-calculating for BULK thermal conductivity suggested that the solder was 15.6 W/m °K, and the worst epoxy was 3.2 W/m °K. This study also monitored the thermal resistance as a function of storage temperature for 150 °C/1000 hr. duration.

With the exception of solder, the data proposed that thermal resistance constantly increases during burn-in of the epoxy in packages. It is suggested that burn-in causes the epoxy to outgass hydrocarbon materials, which will contribute to pin-holes and voiding in the bond-line therefore impeding thermal transfer. The study finally monitored the effect of BLT on thermal resistance.

All six epoxy samples were attached to the TO headers with controlled BLT of 2 mil, 3 mil and 4 mil. The solder controls were not included in this matrix as they were purchased commercially and were not produced in the laboratory. The results propose that the lowest thermal resistance was attained from the 2 mil BLT, expectedly. All epoxies tested demonstrated this condition. As example, Adhesive A had 0.33 °C/W for 2 mil BLT, 0.36 °C/W for 3 mil BLT and 0.41 °C/W for 4 mil BLT. Presumably, a thinner BLT offers a more direct path for heat transfer. Unfortunately, two of the six epoxies tested had such high thermal resistance under 3 and 4 mil BLT condition, that they were immeasurable.

Case Study #5 differed in a few ways from the examples provided above. First of all, it employed electrically non-conductive epoxy instead of silver-filled. Four commercially available boron-nitride filled epoxies were attained for the study. Second, unlike the above cases, the study made use of epoxy outside of the package. A TO-220AC package with A1 tab was glued to copper heat sink outside the package. Thermal resistance from the package was established for each epoxy, and compared with the expected thermal resistance obtained from the vendor's data sheets.[4]

The TO-220AC package glued to the heat sink was made up of a surface area of 100 mm-sq. The BLT was fixed at 3 mils for all epoxies tested. The lowest θJc obtained was 0.6 °C/W, while the worst epoxy had 1.2 °C/W. After back-calculating for BULK thermal conductivity, the study proposed that the best boron nitride filled epoxy was capable of 1.25 W/m °K, while the worst epoxy was 0.62 W/m °K. The strength of the bond between package and heat sink was initially determined; and then as a function of 1000 Thermal Cycles (-55 °C to 125 °C) and 360 hours of Damp Heat (85 °C / 85% RH).

The results of the study indicated that the best thermally conductive epoxy lowest thermal resistance was not the greatest product during environmental testing. Instead, of the four epoxies tested for thermal resistance, the product ranked #2 in the study for heat transfer was capable of out-performing all others during environmental screening. The results of the thermal and environmental tests were needed in order to determine which product would be selected for mounting a heat-sink on top of a microprocessor chip.

Summary of Results

Case Study Package Format Product ID Die Size or square Area of heat path Bond-Line Thickness (BLT) or length of heat path
(units in mils)
Thermal Resistance θJc
(°C / W)
Back-calculated Thermal Conductivity
(W / m°K)
#1 TO-18 solder 20 mil x 20 mil 2 4.8 - 5.3 41
TO-18 Ag epoxy 20 mil x 20 mil 2 6.7 - 7.0 29
#2 TO-5 solder 75 mil x 80 mil 0.5 9.0 -10.3 0.36
TO-5 Ag epoxy 75 mil x 80 mil 0.5 11.3-12.6 0.27
#3 TO-3 A 120 mil x 140 mil 0.37 1.52 0.56
TO-3 B 120 mil x 140 mil 0.37 2.43 0.35
TO-3 C 120 mil x 140 mil 0.34 0.99 0.81
TO-3 D 120 mil x 140 mil 0.45 2.7 0.38
TO-3 E 120 mil x 140 mil 0.68 1.37 1.14
TO-3 F 120 mil x 140 mil 0.25 1.52 0.38
TO-3 G 120 mil x 140 mil 1.4 4.9 0.66
TO-3 solder 120 mil x 140 mil 1 0.9 2.56
#4 TO-3 A 120 mil x 140 mil 2 0.33 14.2
TO-3 B 120 mil x 140 mil 2 0.31 15.1
TO-3 C 120 mil x 140 mil 2 0.4 11.7
TO-3 D 120 mil x 140 mil 2 0.5 9.4
TO-3 E 120 mil x 140 mil 2 1.45 3.2
TO-3 F 120 mil x 140 mil 2 1.38 3.4
TO-3 solder 120 mil x 140 mil 2 0.3 15.6
#5 TO-220AC A 100 mm-sq 3 0.7 -1.0 1.07
T0-220AC B 100 mm-sq 3 0.6 - 0.9 1.25
T0-220AC C 100 mm-sq 3 0.8-1.0 0.94
TO-220AC D 100 mm-sq 3 1.2-1.6 0.62

Conclusions

  • The package format and design make a big difference in thermal resistance and back-calculated BULK thermal conductivity (Case study # 3 versus #4).
  • Solder attachment of chips is always more cool (better thermal management) than epoxy attached ICs (all cases).
  • Epoxy attached chips are about 2° more hot than soldered chips (Case study #1 and #2).
  • Thermal conductivity of silver epoxy can almost be equal to that of solder (Case study #4 and Case study #3) by drawing a comparison with the respective thermal resistance from package.
  • The thinner the bond-line thickness (BLT), the lower the thermal resistance (Case study #1 and #4) and the better BULK thermal conductivity.
  • Boron Nitride filled epoxy demonstrated BULK thermal conductivity values similar with silver epoxy in other packages (Case study #5) and can be an efficient means of thermal management.

References

1. C. Mitchell and H. Berg, Use of Conductive Epoxy for Die Attach, International Microelectronics Symposium (1974), pp. 52 - 58.

2. R. H. Estes, F. W. Kulesza, and C. E. Banfield, Recent Advances Made in Die-Attach Adhesives for Microelectronic Applications, Proceedings of the International Symposium of Hybrid Microelectronics (1985), Anaheim, CA, pp. 391 - 401.

3. R. H. Estes and R. F. Pernice, Die Attach Adhesives Evaluation of VceSAT and 6Jc Performance in Power Devices, Proceedings of the International Symposium on Microelectronics (1989), Baltimore, MD, pp. 664 - 669.

4. M. J. Hodgin and R. H. Estes, Advanced Boron Nitride Epoxy Formulations Excel in Thermal Management Applications, Proceedings of the NEPCON WEST Technical Program (1999), Anaheim, CA, pp. 359 - 366.

Epoxy Technology, Inc

This information has been sourced, reviewed and adapted from materials provided by Epoxy Technology, Inc.

For more information on this source, please visit Epoxy Technology, Inc.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    Epoxy Technology, Inc.. (2019, January 18). Thermal Management with Conductive Adhesives. AZoM. Retrieved on July 07, 2020 from https://www.azom.com/article.aspx?ArticleID=14662.

  • MLA

    Epoxy Technology, Inc.. "Thermal Management with Conductive Adhesives". AZoM. 07 July 2020. <https://www.azom.com/article.aspx?ArticleID=14662>.

  • Chicago

    Epoxy Technology, Inc.. "Thermal Management with Conductive Adhesives". AZoM. https://www.azom.com/article.aspx?ArticleID=14662. (accessed July 07, 2020).

  • Harvard

    Epoxy Technology, Inc.. 2019. Thermal Management with Conductive Adhesives. AZoM, viewed 07 July 2020, https://www.azom.com/article.aspx?ArticleID=14662.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this article?

Leave your feedback
Submit